Emily GriséAhmed El-Geneidy

School of Urban Planning, McGill University

INTRODUCTION

STUDY CONTEXT AND DATA

ACKNOWLEDGEMENTSThe authors wish to acknowledge Marie-Pier Veillette, Robbin Deboosere, and Kevin Manaugh for their efforts in designing the survey, and Daniel Schwartz from IT Customer Services for his assistance in managing the survey distribution. We would also like to thank the three anonymous reviewers for their feedback. Lastly, this work was supported by research grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Social Sciences and Humanities Research Council (SSHRC).

CONCLUSIONS

Agencies should aim to minimize trips involving 2 or more transfers.

Transferring between high frequency routes does not impact trip satisfaction in the same way as transfers involving low frequency services.

Policy implications:

Areas for future research:

Explore satisfaction with transferring buses according to service frequency.

Study how bus stop design, including features such as heated shelters, impact customer satisfaction levels in other cold cities.

Number of transfers:No statistically significant difference between those who transferred once compared to those who did not transfer was observed.

The odds of satisfaction decline by 32% for those who require two or more transfers.

Mode-specific transfers:Transferring between bus routes, and between a bus and subway negatively impact trip satisfaction.

Interestingly, transferring between subway lines did not show an impact on trip satisfaction.

1 Descriptive statistics 2 Logistic regression models

2

Evaluating satisfaction levels and trip characteristics by number of transfers and mode-specific transfers

Model results:Trips involving 1 transfer compared to 0 transfers have the same odds of being satisfied

The odds of being satisfied drops by 32% when 2 transfers or more are required in a trip compared to 0 transfers

Travel time decreased the odds of satisfaction by 5% for every additional minute spent travelling

Variable Odds Ratio Sig. † 95% Conf. Interval

Personal characteristicsCar ownership 1.21 0.89 1.66Household size 0.97 0.86 1.09Child at home 1.09 0.76 1.57Female 0.73 ** 0.55 0.96Other (ref = male) 1.30 0.23 7.47Age 1.01 * 1.00 1.03High income 1.53 0.88 2.67Medium income (ref = low income) 1.48 * 0.99 2.21Trip characteristicsFall trip 1.38 ** 1.06 1.79Downtown campus 2.44 0.79 7.55Travel during peak hour 0.94 0.70 1.28Travel time (min) 0.95 *** 0.93 0.97Travel time squared 1.01 ** 1.00 1.01Type of transferBus-bus transfer 0.63 ** 0.43 0.92Subway-subway transfer 1.03 0.76 1.40Bus-subway transfer 0.73 ** 0.54 0.97Bus-train transfer 1.12 0.52 2.44Train-subway transfer 0.63 0.31 1.26Constant 3.94 ** 0.98 15.81AIC 1445.84BIC 1544.68Log likelihood -703.92Observations 1,342

Satisfaction with last trip with mode-specific controls

†*** Significant at 99% ** Significant at 95% * Significant at 90%, blank cell indicates no statistical significance

Satisfaction with last trip modelVariable Odds Ratio Sig. † 95% Conf. interval

Personal characteristicsCar ownership 1.25 0.92 1.71Household size 0.97 0.86 1.09Child at home 1.08 0.76 1.55Female 0.72 ** 0.54 0.95Other (ref = male) 1.42 0.24 8.38Age 1.01 * 1.00 1.03High income 1.55 0.89 2.69Medium income (ref = low income) 1.48 * 0.99 2.21Trip characteristicsFall trip 1.43 ** 1.10 1.85Downtown campus 2.53 * 0.83 7.66Travel during peak hour 0.97 0.72 1.31Travel time (minutes) 0.95 *** 0.93 0.97Travel time squared 1.01 *** 1.00 1.01Number of transfers (ref = 0 transfers)One transfer 1.02 0.75 1.39Two or more transfers 0.68 ** 0.49 0.96Constant 3.77 * 0.95 14.98AIC 1445.329 BIC 1528.56Log likelihood -706.66Observations 1,342

In the case of a statistically significant difference, the level of significance is represented as follows:*** Significant at 99% ** Significant at 95% * Significant at 90%

Statistical significance of difference in mean satisfaction levels of trips, using a Chi-square test

Number of transfers0 transfers 1 transfer 2 or more transfers

0 transfers ---1 transfer 0.23** ---2 or more transfers 0.7*** 0.24** ---

Trips with zero transfersTrain Bus Subway

Train ---Bus 0.21* ---Subway 0.08 0.29* ---

Mode-specific transfers

Bus-busSubway-subway

Bus-subway Bus-train Train-subway

Bus-bus ---Subway-subway 0.40*** ---Bus-subway 0.25** 0.15 ---Bus-train 0.12 0.28 0.13 ---Train-subway 0.21 0.19 0.04* 0.09 ---

NAverage

satisfaction Average travel

time (min)Average trip distance (km)

Comparing trip details by number of transfers

0 transfer 598 4.13 49.47 12.261 transfer 433 3.90 58.61 14.012 or more transfers 311 3.20 79.66 17.59All respondents 1,342 3.44 72.06 16.27

Trips with zero transfers

Train only 144 4.17 71.77 23.15

Bus only 247 3.96 40.69 7.12

Subway only 207 4.25 35.94 6.53Mode-specific transfers

Bus-bus 154 3.48 62.24 11.47Subway-subway 425 3.88 53.08 11.26Bus-subway 414 3.73 58.91 12.93Bus-train 40 3.60 87.00 26.12Train-subway 51 3.69 77.65 25.32

Satisfaction decreases with number of transfers

Differences in satisfaction levels are observed when examining trends in average satisfaction according to mode-specific types of transfers

Model results:A transfer between 2 bus routes decreases the odds of satisfaction by 37%, compared to a non-transferring trip

Transferring subway lines has no statistically significant impact on trip satisfaction

Commuters who transferred from a subway to a bus or vice versa have 27% lower odds of satisfaction compared to their non-transferring counterparts

ANALYSIS

The data for this study are derived from the 2017/18 McGill University Travel Survey, which include:

Detailed trip characteristics, including number of bus routes, train lines and subway lines used to complete their last trip

Overall trip satisfaction Personal characteristics

´Projection: NAD 1983 MTM 8

Data: Montreal Public Transport Operators & Statistics Canada

0 105 KmMcGill University

Macdonald Campus Subway network

Train network

Bus networkIsland of Montreal

Subway and rail stations

Montreal Public Transport Network

INTRODUCTIONTransferring is strongly associated with trip satisfaction however the model results indicate that the relationship varies according to the number of transfers and the mode(s) being used in a trip.

Conventional wisdom in public transport planning suggests that transfers should be minimized due to negative perceptions associated with them.

However, little scholarly attention has been paid to the association between number and type of transfers and overall satisfaction with public transport services.

The aim of this study is to answer the following three research questions:

(1) Are people that require transfers on their daily commute less satisfied with their trips compared to their non-transferring counterparts?

(2) How many transfers appear to be too many transfers to remain satisfied with a trip?

(3) Do mode-specific transfers have differential impacts on overall satisfaction levels?

An analysis of the influence of transfers on trip satisfactionTransportation Research at McGillRAMTTransferring matters: